Path: utzoo!mnetor!uunet!lll-winken!lll-tis!ames!pasteur!ucbvax!LAVALVM1.BITNET!STAR
From: STAR@LAVALVM1.BITNET (Spencer Star)
Newsgroups: comp.ai.digest
Subject: Re: AIList V6 #45 - Logic, RuleC, Methodology, Constraint
 Languages
Message-ID: <8803050700.AA01647@ucbvax.Berkeley.EDU>
Date: 5 Mar 88 07:44:54 GMT
References: <AIList-REQUEST@SRI.COM>
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 5
Approved: ailist@kl.sri.com
X-Unparsable-Date: Fri, 04 Mar 88 10:48:30 HNE


Paul Creelman notes in the Feb 28th AIList that Speigelhalter says that
probabilities are not the right way to represent uncertainty in expert
systems, and then goes on to list a number of reasons why.  I found this
somewhat surprising since Speigelhalter uses probabilities in his work
on uncertainty in expert systems.  Where does he make all these nasty
comments.